Hypothermic Modulation of Cerebral Ischemic Injury during Cardiopulmonary Bypass in Pigs 

Animals were handled according to the guidelines approved by the American Physiological Society and the “Guide for the Care and Use of Laboratory Animals” published by the National Institutes of Health (publication 85–23, revised 1985). Our institutional animal care and use committee approved this study.

General anesthesia was induced using ketamine (20 mg/kg) given intramuscularly to 41 mature, female Yorkshire pigs weighing 29.9 +/- 1 kg (range, 26–36 kg). After endotracheal intubation, each animal was ventilated with a tidal volume of 15 ml/kg using a volume-cycled ventilator (model 607; Harvard Apparatus, Na-tick, MA) with an air and oxygen mixture. Anesthesia was maintained with a continuous infusion of fentanyl (10 micro gram [center dot] kg sup -1 [center dot] h sup -1) and diazepam (0.3 mg [center dot] kg sup -1 [center dot] h sup -1) through a catheter in a lateral auricular vein. Supplemental boluses of fentanyl (10 micro gram/kg) and diazepam (0.1 mg/kg) or isoflurane (0.25–0.5% inspired concentration) were administered as necessary to maintain a sufficient depth of anesthesia as indicated by eye reflexes and acute increases in blood pressure and heart rate during spontaneous circulation, or by hypertension and cyclic variability in blood pressure during CPB. Muscle paralysis was provided by 0.1 mg/kg pancuronium given intravenously and repeated as required to prevent shivering.

Fluid-filled catheters were inserted in the femoral artery for microsphere sampling and arterial blood gas analysis and in the femoral vein for fluid administration. Core temperature was recorded using a precalibrated thermistor probe (Yellow Springs Instruments, Yellow Springs, OH) placed in the distal esophagus 40 cm from the snout. A pulmonary artery catheter was placed via the right cephalic vein for thermodilution cardiac output measurements before CPB. The pig was turned prone, and both temporalis muscles were dissected free from the overlying cranium and reflected laterally. Along the midline suture, a 1-cm burr hole was drilled to expose the superior sagittal sinus, which was cannulated. A precalibrated thermistor probe, 1-cm long, was inserted directly into the left midcortex to measure brain temperature. Through a separate 0.5-cm burr hole made 1 cm lateral to the midline suture and 0.5 cm caudal to the frontal parietal suture, a microdialysis catheter (model MF-5144; BAS/Carnegie Medicine, West Lafayette, IN) was inserted 6 mm into the lateral gyrus of the right cortex. The position of the dialysis catheter was secured using a plastic harness. The temporalis muscles were reapproximated, and the wound was closed. Two electroencephalographic (EEG) signals were obtained using subcutaneous platinum needle electrodes with the reference electrode placed anterior to the bregma and the active electrode overlying the midparietal region of the cortex (model MP 100; Bio Pac Systems, Goleta, CA). The signals were recorded at a bandwidth of 0.3–30 Hz and stored on computer for later analysis. The animal was turned in the supine position.

Through a median sternotomy incision, transducer-tipped catheters (model MPC 500; Millar Instruments, Houston, TX) were placed in the aorta via the right internal mammary artery and in the superior vena cava via the right internal mammary vein. The innominate and left subclavian arteries were isolated by blunt dissection and encircled with loose suture ligatures. A triple-lumen catheter was inserted through the right superior pulmonary vein and positioned with the proximal lumen in the left atrium. The pericardium was incised and tented. Heparin (300 U/kg) was administered before inserting a 20-French infusion cannula in the ascending aortic arch and a 37-French two-stage venous drainage cannula through the right atrium to harvest venous return. Heparin doses were repeated during CPB as necessary to maintain an activated clotting time > 400 s. The oxygenator was primed with 1,000 ml crystalloid solution (Plasmalyte; Baxter Edwards Critical Care, Deerfield, IL) and 500 ml 6% hetastarch (Hespan; Dupont Pharmaceuticals, Wilmington, DE). A 40-micro meter arterial blood filter and bubble trap (model SP3840; Pall Biomedical Products, East Hills, NY) was inserted in the arterial infusion line, and a membrane oxygenator with reservoir unit (model VPCML Plus; Cobe Cardiovascular, Arvada, CO) was used. Nonpulsatile perfusion was provided by a roller pump (model 5000; Sarns, Ann Arbor, MI). During hypothermic CPB, the arterial carbon dioxide tension was regulated using an alpha-stat technique.

(Figure 1) illustrates the experimental protocol. Before bypass, animals were randomly assigned to one of four temperature groups as measured by the esophageal thermistor: 37 [degree sign] Celsius (n = 10), 34 [degree sign] Celsius (n = 10), 31 [degree sign] Celsius (n = 11), or 28 [degree sign] Celsius (n = 10). Prebypass hemodilution was achieved by placing each animal on CPB for 1–2 min, followed by discontinuation of CPB and temporary removal of the right atrial venous drainage cannula. After stabilizing for 30 min, baseline hemodynamic and blood flow measurements were acquired. Next the venous drainage cannula was reinserted, and the animal was placed on total CPB. After the animal had stabilized for 30 min on normothermic CPB, repeated measurements were obtained (warm CPB interval). Next the animal's esophageal temperature was maintained at normothermia or reduced using a thermostatically controlled water bath (Sams), and measurements were acquired 20 min after stable temperature conditions had been reached (cold CPB interval). The animal was then subjected to 20 min of global cerebral ischemia by temporarily ligating the innominate and left subclavian arteries. During ischemia, pump flow was adjusted to maintain mean aortic pressure measured in the femoral artery at 80 mmHg. The animal was then reperfused for 20 min, followed by systemic rewarming. During the cooling and rewarming phases, the temperature gradient between the water bath and core (esophageal) sites was maintained less than 10 [degree sign] Celsius. The duration of active cooling and rewarming was recorded so that the rates of temperature change (expressed as [degree sign] Celsius/min) could be calculated. Repeated blood flow measurements were obtained during ischemia and reperfusion and after rewarming. Final EEG measurements were acquired 30 min after terminating CPB.

Central venous, sagittal sinus, and aortic pressures were measured using a Gould ES2000 electrostatic recorder (Valley View, OH). Arterial and sagittal sinus blood gases (model 1306; Instrumentation Laboratory, Lexington, MA) and serum hemoglobin (CO-Oximeter model 482, Instrumentation Laboratory) were repeatedly measured. CO-Oximeter saturations were calibrated for animal hemoglobin. After stabilizing on CPB, plasma glucose was determined by a glucometer (Lifescan, Milpitas, CA), and plasma osmolality was measured using a vapor pressure osmometer (Wescor, Logan, UT).

Electroencephalographic signals were analyzed using a five-point visual inspection scale used previously to correlate the EEG with cerebral ischemia. [28,30]Each recording was independently reviewed by two investigators blinded to the animal's temperature group and assigned a numeric score where 5 = normal EEG, indistinguishable from baseline EEG; 4 = mildly (25%) depressed amplitude and frequency from baseline; 3 = moderately (50%) depressed amplitude and frequency from baseline; 2 = markedly (75%) depressed amplitude and frequency from baseline; and 1 = isoelectric EEG. Interobserver reliability of the EEG analysis score was tested using a weighted kappa statistic of the two observers.

Before insertion of the microdialysis catheters, the relative recovery rate of each probe was determined by in vitro dialysis against a solution of known concentration of glucose (2 mg/ml) at a temperature used during the experiment. Preinsertion glucose recovery rates in the microdialysis catheters were not affected by ambient temperature; these rates were 15.2 +/- 0.8% at 28 [degree sign] Celsius, 13.9 +/- 0.6% at 31 [degree sign] Celsius, 14.6 +/- 0.9% at 34 [degree sign] Celsius, and 15.4 +/- 0.9% at 37 [degree sign] Celsius (P = NS between groups). During the experiment, artificial CSF (Na sup + 147 mM, K sup + 4 mm, Mg sup ++ 0.9 mM, Ca sup ++ 2.3 mM, Cl sup - 157 mM in deionized water) was continually pumped at 2 micro liter/min through the 4-mm length of dialysis tubing (diameter, 500 micro meter; molecular weight cutoff, 20,000 Daltons). The effluent was collected for 20-min sampling periods and analyzed by blinded observers to determine the brain's extracellular EAA concentration using high-performance liquid chromatography with naphthaldicarboxaldehyde derivitization for amino acid separation and quantification. [31] 

After terminating CPB, a CSF sample was aspirated via a needle inserted in the cisterna magna and subsequently analyzed for S-100 protein content using a monoclonal two-site immunoradiometric assay (Sangtec 100; Sangtec Medical AB, Bromma, Sweden). This technique uses three monoclonal antibodies to detect the S-100 beta-beta and alpha-beta dimers, which are isoforms specific for astroglial cells. [32]Samples were analyzed in duplicate and rejected if > 10% variability occurred.

At the end of each experiment, the dorsal cerebral cortex surrounding the insertion site of the microdialysis catheter was biopsied using a high-speed suction device that deposited 150–250 mg brain tissue into liquid nitrogen within 0.5 s. [33]The tissue sample was analyzed for protein content using the Lowry modification of the Folin phenol reagent method. [34]The supernatant was filtered and analyzed for high-energy phosphate glucose and lactate. Adenosine triphosphate, adenosine diphosphate, and adenosine monophosphate concentrations were measured by reverse-phase high-performance liquid chromatography. Lactate concentration was enzymatically determined using a spectrophotometer. [33] 

Regional blood flow was measured by injecting 2–3 million radioactive microspheres (15 +/- 3 micro meter in diameter) labeled with strontium-85, cobalt-57, scandium-46, niobium-95, tin-113, or chromium-51, as previously described. [28,33]At baseline, microspheres were injected into the left atrium; during CPB, microspheres were injected directly into the aortic cannula. The order of microspheres was randomized and each dose was selected to ensure that all regional brain tissue samples contained > 400 microspheres. At the conclusion of each experiment, the heart was arrested using an intravenous injection of saturated potassium solution and the animal was exsanguinated. Coronal sections were acquired from each kidney and the radioactivity of the renal cortex was assessed. The brain was removed and total radioactivity was counted from the left and right cerebral hemispheres as well as from the cerebellum, hippocampus, caudate nucleus, brain stem, and spinal cord. Radioactivity was determined in each specimen using a Packard gamma counter (Meriden, CT) with a 3-inch sodium iodide crystal; regional tissue blood flows (ml [center dot] min sup -1 [center dot] 100 g sup -1) were calculated using standard methods. [35]The cerebral metabolic rate for oxygen (CMRO²; ml [center dot] min sup -1 [center dot] 100 g sup -1) was calculated as [(arterial oxygen content - sagittal sinus oxygen content)(cerebral blood flow)][division sign] 100. Cerebral oxygen extraction (%) was calculated as [(arterial oxygen content - sagittal sinus oxygen content)[division sign] arterial oxygen content] x 100.

Outcome variables for repeated measures, such as regional blood flows, hemodynamic variables, EAA neurotransmitters, and EEG scores were analyzed using analysis of variance for a two-factor experiment (temperature and time) with repeated measures over time. Other outcome variables without repeated measures (i.e., S-100 protein and brain metabolites) were analyzed using the Kruskal-Wallis test. Fisher's least-significant difference procedure was used for multiple comparisons, with Bonferroni adjustment. Data are expressed as means +/- SEM; P < 0.05 indicated significance. In addition, EEG data are presented as median and interquartile ranges. To ensure adequate microsphere mixing, paired organ blood flows in the kidneys and cerebral hemispheres were compared by linear regression analyses.

The study was successfully completed in 40 of 41 animals. One animal in the 31 [degree sign] Celsius group was excluded because of excessive intracerebral bleeding after inserting the microdialysis probe and subsequent heparinization. The duration of CPB was similar in all groups, averaging 149 +/- 2 min (range, 140–155 min). In the 34 [degree sign] Celsius, 31 [degree sign] Celsius, and 28 [degree sign] Celsius groups, the rates of core cooling were similar at 0.72 +/- 0.08 [degree sign] Celsius/min, 0.65 +/- 0.07 [degree sign] Celsius/min, and 0.67 +/- 0.05 [degree sign] Celsius/min, respectively. The rates of core rewarming were also similar in the 34 [degree sign] Celsius (0.64 +/- 0.05 [degree sign] Celsius/min), 31 [degree sign] Celsius (0.62 +/- 0.05 [degree sign] Celsius/min), and 28 [degree sign] Celsius (0.67 +/- 0.04 [degree sign] Celsius) groups. There were no intergroup differences in the doses of fentanyl, diazepam, and isoflurane used during the experimental procedure. Central venous pressure, as measured by the mammary vein catheter, remained near zero throughout CPB. Values for arterial pH, carbon dioxide tension, and oxygen tension remained comparable among the groups throughout the procedure.

Hemodynamic and blood gas data acquired during the experimental procedure are shown in Table 1. By design, aortic blood pressure was maintained constant in all groups throughout CPB by adjusting pump flow. At baseline, plasma hemoglobin was comparably reduced without intergroup differences throughout the experiment. Sagittal sinus pressure decreased in all groups after initiation of CPB and during cerebral ischemia. During reperfusion, sagittal sinus pressure increased in all groups but was significantly greater in the 37 [degree sign] Celsius animals. Serum glucose increased during ischemia in the 31 [degree sign] Celsius, 34 [degree sign] Celsius, and 37 [degree sign] Celsius groups and was greater in the 37 [degree sign] Celsius group than in the 28 [degree sign] Celsius animals. As shown in Table 1, the temperature protocol was precisely followed. During ischemia, brain temperature changed slightly but significantly, with a decrease in the 37 [degree sign] Celsius group and an increase in the 28 [degree sign] Celsius group. Brain temperatures were restored after rewarming, although brain temperature in the 28 [degree sign] Celsius animals remained significantly lower than in the 31 [degree sign] Celsius and 34 [degree sign] Celsius groups.

Adequate microsphere mixing during the study was confirmed by comparing paired regional blood flows in the cerebral hemispheres (right cerebral blood flow = 0.98 left cerebral blood flow + 1.76 ml [center dot] min sup -1 [center dot] 100 g sup -1; R = 0.978; P < 0.01) and kidneys (right renal blood flow = 0.97 left renal blood flow + 2.85 ml [center dot] min sup -1 [center dot] 100 g sup -1; R = 0.983; P < 0.01). Cerebral blood flow and CMRO²decreased by 15–25% in all groups with initiation of CPB (Table 2). Subsequent cooling to 34 [degree sign] Celsius, 31 [degree sign] Celsius, or 28 [degree sign] Celsius reduced cerebral perfusion and metabolism further so that significant intergroup differences existed when compared with 37 [degree sign] Celsius animals. Global ischemia to all areas of the brain was achieved by innominate and left subclavian artery ligation, which caused an 80–85% reduction in cerebral blood flow; cerebral blood flow was restored during reperfusion. Similar changes were found throughout other brain regions (Table 2). The CMRO²increased during reperfusion in the 37 [degree sign] Celsius group and was greater than in the other temperature groups. However, after rewarming, CMRO²increased in the 28 [degree sign] Celsius, 31 [degree sign] Celsius, and 34 [degree sign] Celsius groups but not in the 37 [degree sign] Celsius animals so that CMRO²recovery was significantly lower with 37 [degree sign] Celsius than with 28 [degree sign] Celsius CPB (Table 2). Cerebral oxygen extraction did not change with warm CPB and was significantly lower after cooling to 28 [degree sign] Celsius than with the other temperatures. After rewarming, cerebral oxygen extraction increased in the 28 [degree sign] Celsius, 31 [degree sign] Celsius, and 34 [degree sign] Celsius groups and was significantly greater in the 28 [degree sign] Celsius animals than in the 31 [degree sign] Celsius and 34 [degree sign] Celsius animals. In contrast, oxygen extraction remained lower in the 37 [degree sign] Celsius group than the other three temperature groups.

The kappa statistic comparing the two EEG graders was 0.58, indicating adequate interobserver agreement. All animals had comparable EEG scores at baseline after hemodilution, which did not change during warm CPB (Table 3). Cooling to 28 [degree sign] Celsius reduced the EEG score, making it significantly lower than in the 34 [degree sign] Celsius and 37 [degree sign] Celsius groups. The time to isoelectric EEG with ischemia was longer in the 28 [degree sign] Celsius (28.3 +/- 3.9 s) and 31 [degree sign] Celsius (23.2 +/- 2.1 s) groups than in the 34 [degree sign] Celsius (14.7 +/- 1.5 s) and 37 [degree sign] Celsius (13.4 +/- 1.0 s) animals (P < 0.05, respectively). Electroencephalogram scores significantly recovered during reperfusion and rewarming in the 28 [degree sign] Celsius, 31 [degree sign] Celsius, and 34 [degree sign] Celsius animals but not in the 37 [degree sign] Celsius group (Table 3). With reperfusion, the time to return of EEG activity was 60–70% longer in the 37 [degree sign] Celsius group than the other temperature groups (P < 0.05, respectively). Thirty minutes after terminating CPB, EEG scores were depressed in all groups when compared with baseline values but were significantly lower in the 37 [degree sign] Celsius animals than in the other three temperature groups. Further, EEG scores off CPB were greater in the 28 [degree sign] Celsius animals than in the 34 [degree sign] Celsius group.

Brain glutamate (Figure 2(A)) and aspartate (Figure 2(B)) did not change with ischemia, reperfusion, or rewarming in the 28 [degree sign] Celsius and 31 [degree sign] Celsius groups. In contrast, glutamate increased nearly six times during ischemia in the 37 [degree sign] Celsius group and was significantly greater during reperfusion in the 34 [degree sign] Celsius and 37 [degree sign] Celsius groups compared with the 28 [degree sign] Celsius and 31 [degree sign] Celsius animals. Figure 3shows data from all animals comparing individual brain temperatures at the time of ischemia with glutamate release during reperfusion. Overall, there was a direct relation between brain temperature and log glutamate concentration during ischemia (R = 0.642; P < 0.01) and during reperfusion (R = 0.653; P < 0.01). Similarly, aspartate increased during ischemia in the 34 [degree sign] Celsius and 37 [degree sign] Celsius groups and remained significantly greater than in the 28 [degree sign] Celsius group during reperfusion (Figure 1(B)). After rewarming and terminating CPB, glutamate and aspartate concentrations returned to baseline in all groups.

Cerebrospinal fluid concentration of S-100 protein was higher in the 34 [degree sign] Celsius animals than the 28 [degree sign] Celsius and 31 [degree sign] Celsius groups and was greater in the 37 [degree sign] Celsius animals than all other temperature groups (Table 4). The concentration of S-100 protein in the CSF after CPB correlated directly with brain log glutamate release during reperfusion (R = 0.507; P < 0.02). In the cortical brain biopsy samples (Table 4), there were no intergroup differences in adenosine monophosphate, adenosine diphosphate, and adenosine triphosphate concentrations, although lactate was two to three times greater in the 37 [degree sign] Celsius animals compared with the other groups.

These data clearly indicate the dangers inherent in transient cerebral ischemia during 34 [degree sign] Celsius CPB and particularly during 37 [degree sign] Celsius CPB. In contrast, cooling to 31 [degree sign] Celsius and 28 [degree sign] Celsius before ischemia significantly improved the recovery of acute indices of cerebral injury. Significant release of glutamate and aspartate was found in normothermic (37 [degree sign] Celsius) brains and in those cooled to 34 [degree sign] Celsius. In contrast, systemic cooling before cerebral ischemia to 28 [degree sign] Celsius or 31 [degree sign] Celsius proved equally beneficial in terms of attenuating EAA release and preserving metabolic recovery. In all animals, acute neurologic morbidity, as assessed by EEG, cerebral metabolism, and S-100 protein concentrations in the CSF, correlated directly with brain intracellular glutamate released during ischemia and reperfusion. In this CPB model with 20 min of global cerebral ischemia, there was no greater additional benefit with cooling to 28 [degree sign] Celsius than to 31 [degree sign] Celsius, although differences may occur with more severe levels of ischemia.

This study examined potential neuroprotective strategies that could attenuate excitatory neurotransmitter release, suppress cerebral oxygen metabolism, reduce the systemic glucose response to cerebral ischemia, or enhance collateral cerebral perfusion. In these CPB experiments, significant effects from hypothermia were found by the first three mechanisms. In addition, other studies have found that hypothermia provides neuroprotection by inhibiting hypoxic brain depolarization and protease activation, [36]mitigating edema and leukotriene formation, [37]delaying ionic calcium influx, [38]and decreasing hydroxyl radical formation during reperfusion. [36] 

During cerebral ischemia, release of glutamate and aspartate into the synaptic cleft can increase EAA levels 25 to 100 times higher than normal. As a result, intracellular calcium increases markedly and activates kinases, endonucleases, and phospholipases, which, in turn, cause neuronal injury and death. [39]A direct linear correlation has been established between the amount of EAA released during ischemia and subsequent brain infarct volume in some [40]but not all [41]studies. In the present experiment, a temperature-dependent effect on the release of EAA, particularly glutamate, was found during ischemia and reperfusion. Busto et al. [22]established the importance of directly measuring brain temperature because temporary normothermic cerebral ischemia in the rat reduced brain temperature by 5–6 [degree sign] Celsius, which by itself had a profound protective effect on ischemia-induced glutamate release. [23]The minor temperature fluctuations in the present study most likely reflect the larger size of the pig cranium compared with that of the rat.

However, even when brain temperatures were maintained constant, several studies found that preischemic cooling to 33–35 [degree sign] Celsius completely suppressed extracellular glutamate release [42]and prevented [25]or markedly attenuated [24]histopathologic injury. In contrast, the present study of the pig on CPB found that 31 [degree sign] Celsius but not 34 [degree sign] Celsius improved EEG recovery, brain tissue lactate, and CSF S-100 protein release, and reduced EAA release. The cause of any discrepancy is not known but may be related to differences in periischemic collateral blood flow in the brain when pulsatile and nonpulsatile perfusion techniques are compared. Previous studies have documented 42–55% lower collateral blood flow during cerebral ischemia on CPB than in animals without CPB [28,29]; this lower blood flow in turn worsened EEG recovery. [29]Consequently, the assumption that extracorporeal perfusion at any temperature < 35 [degree sign] Celsius is neuroprotective may not be valid.

Two factors that significantly affect cerebral metabolism during CPB have been identified: the loss of pulsatile flow and temperature. In the present study, initiation of CPB reduced CMRO²in all groups by 20–25% when compared with baseline, hemodiluted values under pulsatile flow conditions. This reduction is similar to that reported previously by us [28]and others [43,44]in experimental animals and in humans [27]and is thought to represent depressed neuronal activity, functional capillary closure, or both. However, as seen in the 37 [degree sign] Celsius animals, the decrement in CMRO²with CPB was not sufficient to attenuate markers of ischemic injury by itself. In contrast, cooling during CPB to 34 [degree sign] Celsius reduced CMRO²from baseline by 45% with nearly a 70% reduction at 31 [degree sign] Celsius and 28 [degree sign] Celsius. Cooling to 31 [degree sign] Celsius reduced CMRO²to levels comparable to those achieved at 28 [degree sign] Celsius, indicating that further cooling below 31 [degree sign] Celsius may be unnecessary. In this model, we could not differentiate the amount of hypothermic neuroprotection attributable to cerebral metabolic depression from suppression of EAA release. Animal studies by Verhaegen et al. [45]and Nakas-hima and Todd [46]found that factors other than metabolic depression from hypothermia are responsible for controlling energy use after ischemia, emphasizing the role of reduced EAA release.

Recent clinical studies have documented greater neurologic damage in terms of cognitive function [47]and central deficits [19]in patients maintained at normothermia during CPB and requiring active rewarming than in patients cooled below 32 [degree sign] Celsius. Our data support this concept because 37 [degree sign] Celsius animals suffered the worst outcome after temporary ischemia. Problems arise when comparing clinical studies of normothermic and hypothermic CPB due to critical differences in serum glucose concentrations, [19]the need for systemic vasopressors, [47]and the duration of CPB, [17]which are all confounding factors that could adversely affect the extent of neurologic injury during CPB. The present study avoided some of these influences because the duration of CPB was similar between groups and vasopressors were not used. Serum glucose was higher during ischemia in the 37 [degree sign] Celsius group than in 28 [degree sing] Celsius animals but not by amounts reported to increase neurologic injury. [48]In the study of Regragui et al., [47]the lack of further benefit as patients were cooled from 32 [degree sign] Celsius to 28 [degree sign] Celsius may be explained by our results, in which similar attenuation of glutamate and aspartate release was found during cerebral ischemia at 31 [degree sign] Celsius and 28 [degree sign] Celsius.

The cause of neurologic damage during CPB is ischemia from cerebral emboli, although there is no gold standard animal model to study this phenomenon during CPB. Our porcine model is not designed to replicate the wide clinical spectrum of cerebral ischemic injury seen in patients that encompasses mild cognitive defects to frank strokes and encephalopathy. Rather, this experimental paradigm incorporates the principal causative factors that occur during CPB, specifically nonpulsatile flow, [28,29]cerebral microemboli, [35]altered cerebral metabolism, [27,28]and hemodilution [49]with acute blood flow deprivation under reproducible and controlled conditions to define the importance of graded hypothermia. As such, there are obvious limitations when extrapolating our findings to the clinical situation. Nevertheless, this study provides a rationale for selecting the level of core cooling that can improve cerebral recovery after 20 min of temporary ischemia during CPB. Under these experimental conditions, preischemic cooling to 28 [degree sign] Celsius and 31 [degree sign] Celsius proved beneficial.

Several potential limitations of this study should be addressed. S-100 protein was measured as a CSF marker of ischemic glial cell damage, although recovery of this protein has not been directly correlated with functional or histopathologic outcome after CPB. Furthermore, results of this study may not be applicable to longer durations of ischemia or to permanent occlusion without reperfusion. Although periods of potential cerebral embolization can be anticipated during cardiac surgery, hypothermia can only be instituted after cannulation and institution of CPB. Any effect from uneven temperature gradients in the brain were reduced by stabilizing at the designated temperature for at least 20 min before ischemia; less equilibration time could reduce any benefit. Hypothermia may appear neuroprotective by temporally slowing the rate of cell death while not preserving long-term function. However, the lack of permanent benefit has been described only for postischemic but not intraischemic hypothermia. [50]Survival studies after cerebral ischemia during CPB would be necessary to prove the long-term benefits of hypothermia. Finally, rewarming injury (as assessed by EAA release) was not seen in this model, which may reflect the neuroprotective properties of 28 [degree sign] Celsius and 31 [degree sign] Celsius hypothermia.

In conclusion, hypothermia to 28 [degree sign] Celsius or 31 [degree sign] Celsius significantly improved the acute recovery of EAA levels, EEG and cerebral metabolism, and CSF S-100 protein levels after 20 min of temporary global ischemia imposed during CPB. In contrast, temporary cerebral ischemia at 34 [degree sign] Celsius and 37 [degree sign] Celsius CPB caused greater EAA release and neurologic morbidity. Cooling to at least 31 [degree sign] Celsius was necessary to improve acute recovery after global ischemia during CPB. Based on these data, this experimental model of transient cerebral ischemia during CPB can serve to evaluate potential therapeutic strategies that may improve cerebral outcome.

The authors thank David McAdoo, Ph.D., of the Marine Biological Institute, for analyzing the excitatory amino acids; Karen Inners-McBride and Gregory Asimakis, Ph.D., for analyzing brain tissue; Deborah Goodyear and Michelle Farrow for preparing the manuscript; and Faith McLellan for editorial review.